Build with Claude Code From zero to deployed apps with AI-assisted development (Taeho Jo) (z-library.sk, 1lib.sk, z-lib.sk)

【One-Line Pitch】 A practical, project-driven introduction to building and shipping software by collaborating with Claude Code—even if you have little traditional programming background. Best for beginners, makers, and non-engineers who want a structured path from idea to deployed app. 【Book Arc】 - **Opening (~0%–14%)**: Sets the stage by surveying the AI coding landscape and introducing "Vibe Coding" through Claude's web interface, then gives a brief history of AI from neural networks to transformers to explain why these tools work (Early). - **Early (~14%–36%)**: Moves from casual prompting to structured prompting—using 5W1H and PRD-style planning to turn vague ideas into precise instructions—then installs Claude Code and runs a first real project (a handwriting/digit recognition program) across desktop and web versions (Early). - **Middle (~36%–50%)**: Deepens project management with CLAUDE.md (project-specific and global configs), the `/init` command, and session/context handling (`--resume`, `/context`, `/compact`), while building a to-do app through step-by-step prompts and iterative design feedback (Middle). - **Late (~50%–75%)**: Expands into AI-powered applications—connecting external APIs, writing PRDs for projects like FridgeChef, and using subagents (code reviewer, optimization, UX design) that collaborate as a small AI development team (Late). - **Ending (~75%–100%)**: Extends Claude Code into a full environment via MCP servers, Skills, and plugins, then deploys database-backed apps using Vercel and Supabase (Ending). *Note: later-chapter detail is inferred from the table of contents and blurb; excerpts do not cover the final deployment chapters in depth.* 【Key Takeaways】 - **Vibe Coding lowers the entry barrier** (Early): You can ask Claude to build in natural language without programming terminology, making artifacts, prompts, publishing, and customization the four core concepts of the first workflow. - **Planning is the real prompt** (Early): A PRD built on 5W1H—especially "What" and "How"—converts fuzzy ideas into actionable instructions; the book treats the product plan and the prompt as fundamentally the same artifact. - **Section IDs enable surgical edits** (Early): Naming each section of a page lets you revise one part without regenerating everything, a small habit with large iteration benefits. - **CLAUDE.md is the project's memory** (Middle): It records structure, rules, and commands, is auto-read by Claude Code, and can be scoped per project or globally in the home directory—use `/init` to bootstrap it. - **Context is a managed resource** (Middle): Sessions persist on disk and can be resumed with `claude --resume`; `/compact` and `/context` keep long projects from drowning in history. - **Subagents simulate a dev team** (Late): Specialized agents for code review, optimization, and UX design can collaborate, turning one assistant into a structured workflow. - **Extensibility is the payoff** (Ending): MCP servers connect external tools and services, while Skills and plugins customize Claude Code into a tailored development environment. - **Deployment closes the loop** (Ending): Database-backed apps shipped via Vercel and Supabase turn exercises into real, shareable products. 【Reading Tips】 - **Skim Chapter 1's AI history** if you already know the basics; spend that time on the prompting and CLAUDE.md chapters, which carry the most transferable value. - **Build alongside the book** using the downloadable code bundle and GitHub repo—projects like the to-do app and FridgeChef are designed to be extended, not just read. - **Treat context management as a skill to practice deliberately**: experiment with `/compact` and `--resume` on a long project rather than glossing over them. - **Expect plan-tier friction**: the book recommends the Max plan or higher for hands-on chapters; check your access before starting. - **Don't skip the review questions**—they reinforce prompt-writing pitfalls that recur throughout later projects. 【Coverage Limits】 This guide is based on stratified excerpts covering roughly the first half of the book plus the table of contents; the MCP, Skills, plugins, and deployment chapters are summarized from headings and blurb rather than read in detail.